Research map / Reinforcement learning / Policy gradient and actor-critic
Counterexample Guided RL Policy Refinement Using Bayesian Optimization
Briti Gangopadhyay, Pallab Dasgupta
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| refine the PPO policy with counterexample-guided updates instead of training from scratch with a violation penalty (Policy A) | PPO · Cart-pole | failures: 179 -> 0 | Improved |
| refine with counterexample-guided updates instead of Policy A | PPO · Pendulum | failures: 89 -> 89 | No clear effect |
| refine with counterexample-guided updates instead of Policy A | PPO · BipedalWalker | failures: 45 -> 0 | Improved |
| refine with counterexample-guided updates instead of Policy A | PPO · LunarLander Continuous | failures: 42 -> 0 | Improved |
| refinement instead of Policy B (scratch training on only counterexample traces) | PPO · Pendulum | failures: 102 -> 89 | Improved |
| refinement instead of Policy B | PPO · BipedalWalker | failures: 145 -> 0 | Improved |
| refinement instead of Policy C (scratch with original + counterexample traces) | PPO · BipedalWalker | failures: 41 -> 0 | Improved |
| refinement instead of Policy C | PPO · LunarLander Continuous | failures: 5 -> 0 | Improved |
| refinement instead of Policy C | PPO · Cart-pole | training steps: 1M -> 150K | Improved |
| refinement instead of Policy C | PPO · BipedalWalker | training steps: 8M -> 2.8M | Improved |
| refinement instead of Policy A | PPO · Pendulum | training steps: 1.6M -> 850K | Improved |
| add penalty term for property violation when training from scratch (Policy A) without exposing parameter variations | PPO · Cart-pole | failures: 179 | Hurt |
Closest papers by meaning
- Safeguarded Progress in Reinforcement Learning: Safe Bayesian Exploration for Control Policy Synthesis (AAAI 2024 · similarity 0.90)
- Probabilistic Shielding for Safe Reinforcement Learning (AAAI 2025 · similarity 0.89)
- Reinforcement Learning with Adaptive Regularization for Safe Control of Critical Systems (NeurIPS 2024 · similarity 0.89)
- A Provable Approach for End-to-End Safe Reinforcement Learning (NeurIPS 2025 · similarity 0.88)
- Embedding Safety into RL: A New Take on Trust Region Methods (ICML 2025 · similarity 0.88)
- Long-Term Safe Reinforcement Learning with Binary Feedback (AAAI 2024 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
